NEURAL NETWORK APPROACH FOR RECONSTRUCTION OF GEOMAGNETIC DATA VIA MACHINE LEARNING

Abstract

The uprightness of geomagnetic information is a basic factor in understanding the trans- formative procedure of Earth's attractive field, as it gives helpful data to approach surface investigation,   unexploded touchy arms identification, thus on. Hence to solve these issues, this study represents the usage of various machine learning models that helps in reconstruction of data and its interpolation into single dimension. Already existing interpolation techniques seems to not have a greater effectiveness. Hence different machine learning model is taken which are SVM, Decision Tree, RNN are used. The Neural network steps in deep learning process which helps to simulate by manipulating the hidden layers. It mentions the hyperplane to be continuous from the data used for training. The models are evaluated and their performance is estimated with LSTM approach. This approach excels compared to the above listed machine learning models and their values are compared. The accuracy seems to differ at a range of 10% between these models.